CN-122022812-A - Payment risk analysis decision method, device and storage medium

CN122022812ACN 122022812 ACN122022812 ACN 122022812ACN-122022812-A

Abstract

The application discloses a payment risk analysis decision method, a payment risk analysis decision device and a storage medium, which are used for improving the identification accuracy of complex risks. The method comprises the steps of integrating internal data and external data to form a three-dimensional data source, carrying out data screening processing and risk feature extraction processing on the three-dimensional data source, carrying out risk recognition on the processed three-dimensional data source to obtain a risk recognition result, wherein the risk feature extraction processing is used for extracting transaction features, merchant behavior features, merchant attribute features and associated features from the screened three-dimensional data source, executing a corresponding disposal strategy according to the risk level in the risk recognition result, wherein the disposal strategy comprises low risk triggering reminding, medium risk triggering manual review and high risk triggering real-time freezing, and converting the execution result of the disposal strategy into specific business protection action to form a complete risk prevention and control closed loop.

Inventors

LIU YAN
Shu Qingzhou
Li Zuorui
LI XIAOWANG
WANG YUN

Assignees

嘉联支付有限公司

Dates

Publication Date: 20260512
Application Date: 20251205

Claims (10)

1. A payment risk analysis decision method, comprising: integrating internal data and external data to form a three-dimensional data source, wherein the internal data is subject behavior and management data provided by a payment service provider, and the external data is risk and compliance data provided by a third party service; Performing data screening processing and risk feature extraction processing on the three-dimensional data sources, and performing risk identification on the three-dimensional data sources after processing to obtain a risk identification result, wherein the risk feature extraction processing is used for extracting transaction features, merchant behavior features, merchant attribute features and associated features from the three-dimensional data sources after screening; executing a corresponding treatment strategy according to the risk level in the risk identification result, wherein the treatment strategy comprises low risk triggering reminding, medium risk triggering manual review and high risk triggering real-time freezing; and converting an execution result of the treatment strategy into a specific service protection action to form a complete risk prevention and control closed loop.
2. The payment risk analysis decision method of claim 1, wherein the performing data screening and feature extraction processing on the stereoscopic data source comprises: Carrying out real-time cleaning, format standardization and abnormal data filtering on the high-frequency transaction data in the three-dimensional data source based on a real-time streaming computing technology, wherein the high-frequency transaction data is transaction data with a transaction frequency higher than a preset frequency generated in real time; And extracting the screened merchant historical transaction data, merchant behavior data, merchant attribute data and merchant group risk labels in the three-dimensional data source based on an offline analysis technology.
3. The payment risk analysis decision method according to claim 1, wherein the risk identification of the processed three-dimensional data source to obtain a risk identification result comprises: constructing a risk analysis model, wherein the risk analysis model takes the processed three-dimensional data source as input, and takes historical risk events and normal service data as training sets for model training, wherein the normal service data is payment service data extracted from the three-dimensional data source and not triggering risk early warning; And analyzing the three-dimensional data source through the trained risk analysis model to generate a risk identification result comprising a risk data identifier, a risk probability value and a risk feature.
4. The payment risk analysis decision process of claim 1, wherein the integrating internal data with external data to form a stereoscopic data source comprises: and after the sensitive information in the internal data and the external data is eliminated through data desensitization processing, establishing an association mapping relation to form a structured three-dimensional data source.
5. The payment risk analysis decision method of claim 1, wherein after processing the stereoscopic data source and performing risk identification on the processed stereoscopic data source to obtain a risk identification result, before executing a corresponding treatment policy according to a risk level in the risk identification result, the method further comprises: acquiring a risk grade in the risk identification result; And demarcating the risk level based on a risk probability value in the risk identification result, wherein when the risk probability value is smaller than a first preset threshold value, the risk level is judged to be low risk, when the risk probability value is larger than or equal to the first preset threshold value and smaller than a second preset threshold value, the risk level is judged to be medium risk, and when the risk probability value is larger than or equal to the second preset threshold value, the risk level is judged to be high risk.
6. The payment risk analysis decision method according to claim 1, wherein the converting the execution result of the treatment policy into a specific service protection action forms a complete risk prevention and control closed loop, and includes: Synchronizing the execution result of the treatment strategy to a payment system, an account system and an air control system in real time through an API gateway; And executing interception or release operation in the payment system, implementing quota adjustment or functional limitation in the account system, and updating the merchant risk portrait in the wind control system.
7. A payment risk analysis decision process according to any one of claims 1 to 6, wherein the process further comprises: and carrying out full-link log record on the execution details of the service protection action and the execution result.
8. A payment risk analysis decision device, the device comprising: The integration unit is used for integrating internal data and external data to form a three-dimensional data source, wherein the internal data is subject behavior and management data provided by a payment service provider, and the external data is risk and compliance data provided by a third party service; the analysis unit is used for carrying out data screening processing and risk feature extraction processing on the three-dimensional data source, carrying out risk identification on the three-dimensional data source after processing to obtain a risk identification result, and extracting transaction features, merchant behavior features, merchant attribute features and associated features from the three-dimensional data source after screening; The execution unit is used for executing a corresponding treatment strategy according to the risk level in the risk identification result, wherein the treatment strategy comprises low risk triggering reminding, medium risk triggering manual review and high risk triggering real-time freezing; and the conversion unit is used for converting the execution result of the treatment strategy into a specific service protection action to form a complete risk prevention and control closed loop.
9. A payment risk analysis decision device, the device comprising: a processor, a memory, an input-output unit, and a bus; The processor is connected with the memory, the input/output unit and the bus; The memory holds a program that the processor invokes to perform the payment risk analysis decision method of any one of claims 1 to 7.
10. A computer readable storage medium having stored thereon a program which, when executed on a computer, performs the payment risk analysis decision method of any of claims 1 to 7.

Description

Payment risk analysis decision method, device and storage medium Technical Field The present application relates to the field of payment technologies, and in particular, to a payment risk analysis decision method, apparatus, and storage medium. Background With the rapid development of digital economy, payment business has been deeply integrated into the full scene of online and offline, becoming a core link of socioeconomic activity. The payment risk prevention and control is used as a key support for guaranteeing transaction safety and maintaining financial order, and the intelligent level of the payment risk prevention and control directly influences the efficiency and reliability of payment service. In the current business scenario, payment risk analysis relies on deep fusion of multidimensional data, namely internal transaction data, and external risk data provides cross-domain risk clues, and cooperative application of the two is the basis for accurately identifying risks such as fraudulent transactions, money laundering behaviors and the like. The existing payment risk analysis decision scheme mainly adopts a mode of 'rule engine leading+manual assistance', and the flow is that firstly, fixed rules are extracted based on historical risk cases, the rules are solidified into a system, secondly, when transaction occurs, the system simply cleans transaction data, then, risk grades are judged through matching preset rules, finally, treatment actions (such as low risk transaction direct release and high risk transaction interception) are executed according to rule matching results, and complex scenes are rechecked by manual intervention. However, the internal transaction data and the external risk data are in a splitting state, the system can only perform rule matching based on local features of a single data source, cross-source hidden risk association is difficult to mine, the miss judgment rate of complex risks is high, a solidified rule system depends on manual continuous updating, the rule adjustment speed is far slower than the risk evolution speed facing to novel risks, the novel risks cannot be intercepted in time, and normal transactions are easy to misjudge due to rule logic of one tool. Disclosure of Invention The application discloses a payment risk analysis decision method, a payment risk analysis decision device and a storage medium, which are used for improving the identification accuracy of complex risks. The first aspect of the application discloses a payment risk analysis decision method, comprising the following steps: integrating internal data and external data to form a three-dimensional data source, wherein the internal data is subject behavior and management data provided by a payment service provider, and the external data is risk and compliance data provided by a third party service; Performing data screening processing and risk feature extraction processing on the three-dimensional data sources, and performing risk identification on the three-dimensional data sources after processing to obtain a risk identification result, wherein the risk feature extraction processing is used for extracting transaction features, merchant behavior features, merchant attribute features and associated features from the three-dimensional data sources after screening; executing a corresponding treatment strategy according to the risk level in the risk identification result, wherein the treatment strategy comprises low risk triggering reminding, medium risk triggering manual review and high risk triggering real-time freezing; and converting an execution result of the treatment strategy into a specific service protection action to form a complete risk prevention and control closed loop. Optionally, the performing data screening processing and feature extraction processing on the stereoscopic data source includes: Carrying out real-time cleaning, format standardization and abnormal data filtering on the high-frequency transaction data in the three-dimensional data source based on a real-time streaming computing technology, wherein the high-frequency transaction data is transaction data with a transaction frequency higher than a preset frequency generated in real time; And extracting the screened merchant historical transaction data, merchant behavior data, merchant attribute data and merchant group risk labels in the three-dimensional data source based on an offline analysis technology. Optionally, the risk identification is performed on the processed three-dimensional data source to obtain a risk identification result, which includes: constructing a risk analysis model, wherein the risk analysis model takes the processed three-dimensional data source as input, and takes historical risk events and normal service data as training sets for model training, wherein the normal service data is payment service data extracted from the three-dimensional data source and not triggering risk early warning; And analyz